1　Materials and Methods

1.1　Site Description and Soils

Eight representative soil samples were selected from Xianning County(29°39'N to 30°02'N,114°06'E to 114°43'E)in the southeastern part of Hubei,China(Table 1).The soil parent materials were quaternary red clay and shale.According to the U.S.Soil Taxonomy Classification,the soils are Ultisols(Yan et al.,2008).Ultisols contain various textural compositions and exhibit different aggregate mechanical properties and water stability.

Soil samples were carefully collected from the surface soil layer(0～15cm)and carried to the laboratory in rigid boxes and immediately air-dried.Large clods were broken down manually when the proper moisture content was reached.The air-dried samples were ground to a maximum particle diameter of 5mm.The aggregates(3-5mm)were obtained by dry sieving the air-dried soil for 24h at 40℃to limit moisture variations before the aggregate tests were conducted.

Bulk densities were estimated using 100-m L cylinders.Soil p H was measured at a soil-to-water suspension ratio of 1∶2.5.Particle size distribution was determined by the standard pipette procedure after the samples were treated with H2 O2 and shaken in Na hexametaphosphate.Soil organic carbon(SOC)content was determined by oxidation with potassium dichromate in a heated oil bath.Cation exchange capacity(CEC)was determined by treatment of the sample with ammonium acetate buffer at p H 7.0.Fe(Ⅲ)and aluminum oxides(Fed and Ald)were extracted using a Na-dithionite-citrate bicarbonate system(Mehra and Jackson,1960).Poorly crystalline oxides(Feo and Alo)were extracted with NH4 oxalate(Mckeague and Day,1966).All of the cations were determined by ICP(VISTA.MPX;Varian,Inc.,Palo Alto,Calif).The important soil characteristics are listed in Table 2.

1.2　Aggregate Stability Tests

1.2.1　Tensile Strength Test

The tensile strength of aggregates was determined by a crushing method(Dexter and Kroesbergen,1985;Dexter and Watts,2001)with an electronically controlled loading frame.Each oven-dried aggregate was weighed and placed between two parallel plates in a stable position.The lower plate was raised at a constant rate of 2mm·min-1 until the aggregates failed.The maximum reading was recorded using an electronic load cell when the aggregate was fractured.Tensile strength was then calculated using the following equation:

where 0.576 is the proportionality constant,P is the pressure at failure(N),and D is the mean diameter of aggregates(4mm).

1.2.2　Aggregate Penetration Resistance Test

The penetration resistance of aggregate(3～5mm)was tested using dry and wet aggregates.The wet aggregates were placed in a sandbox and equilibrated by water at a tension of-0.1k Pa for 3 days.At least 10 aggregates were used for each treatment.The penetration resistance test was conducted using procedures similar to those used for the tensile strength test.A 1mm diameter probe(total cone angle,30 degrees)was supported by three glass spheres and driven by a motor at a rate of 2mm min-1 into the aggregates.The maximum penetration force(Fmax)was measured using a digital electronic balance(Misra et al.,1986).P was calculated from Fmax as follows:

APR=Fmax/πα2

whereαis the probe radius.

1.2.3　Aggregate Water Stability Test

Aggregate water stability was determined by a routine wet sieving method according to Yoder(1936),with slight modifications.A sample of the aggregates(50g)was immersed in water with different sizes of sieves(2,1,0.5,0.25,and 0.105mm)and shaken for 30min at a frequency of 35cycles min-1,with a stroke length of 4cm.The water-stable aggregates retained on the sieves were ovendried and weighed to obtain a stable aggregate mass.The mean weight diameter(MWD)was calculated using the routine method:

where ri is the aperture of the ith mesh in mm,r0=r1,rn=rn+1;mi is the mass fraction remaining on i th sieve;and n is the number of sieves.

1.3　Aggregate Breakdown Induced by Rainfall

The method used to test aggregate breakdown was similar to that used by Legout et al.(2005a).This method has also been used extensively in other studies(Young,1984;Glanville and Smith 1988;Truman et al.,1990).Eighteen aggregate breakdown devices were built using a sieve to hold the aggregates.These devices were then placed in a bucket to retain the splashed fragments inside.The aggregates(5g)were placed on a filter paper in the sieve and then placed in a rainfall simulator at an intensity of 60mm·h-1.The rainfall simulator was composed of a SPRACO cone jet nozzle mounted 4.75m above the surface of the aggregates(Luk et al.,1986).The medium drop size was 2.4mm,with a uniformity coefficient of 90%(Luk et al.,1986;Cai et al.,2005).Before the experiment was conducted,the simulator was calibrated with a uniformity of more than 95%for the locations of the 18 devices.

Our previous experiments revealed that aggregate MWD are constant after 24 min.For each soil type in the present study,the 18 devices with aggregates were subjected to rainfall from 0.5 to 24 min,respectively.After various amounts of rainfall were received by the aggregates,the whole fragments from aggregate breakdown were gently washed into a beaker with ethanol to preserve the structure of the fragments.The fragments were dried in an oven at 40℃,weighed,and gently dry sieved by hand in a column of eight sieves(10,5,2,1,0.5,0.25,0.105,and 0.05mm).Aggregate stability was expressed using MWDrain.The mass of splash was calculated from the difference between original and final values of oven-dry mass.The splash loss of the 24-mm cumulative rainfall event was considered to be the total splash.

1.4　Data Analyses

The treatments for each soil type were replicated three times.Analyses were performed in SPSS 18.0.Significant differences among the treatments of soil,aggregate mechanical properties,and aggregate water stability were determined by the least significant difference test for a multiple range test at the 0.05 significance level.Correlation coefficients were based on the means of the replicates of eight soil samples.